摘要
为了获取用户的信息需求,并依据信息需求模型在因特网上搜索相关文本,文章提出了基于示例的用户信息需求模型的获取和表示方法。其基本思想是:在用户给定的示例文本集的基础上,利用特征项的类别区分度,抽取能够表现用户兴趣的项作为用户信息需求模型的基本特征项集。然后,基于统计上的Fisher准则,进行判别分析,以获取特征项在相关文本的判定中的重要程度。最后,给出用户信息需求模型的逻辑和物理表示。
The paper presents an approach for users' profiles acquisition and representation based on example texts in order to acquire the users' information requirements and seek the related texts according to them.Its main idea is shown as follows:Based on the example texts given by users,it applies the term class discrimination to extract the basic terms to representation of users' interests,and then it performs the discriminant analysis by Fisher Rule to obtain the terms' weights in discriminating the related texts.Finally,it gives the logic and physical representation for users' pro- files.
出处
《计算机工程与应用》
CSCD
北大核心
2000年第9期11-12,16,共3页
Computer Engineering and Applications
基金
国家自然科学基金资助项目!(编号:69675019)
国家教委博士点基金资助项目。
关键词
文本过滤
用户信息需求模型
示例
INTERNET网
Text Filtering, User Profiles,Text Feature Extraction, Vector Space Model, Discriminant Analysis